This chapter looks at one important aspect of the syntax of 'want': whether the notional subject of its complement predication is left implicit when it is coreferential with the wanter, or whether it is expressed overtly in the complement clause by means of a pronominal element.
The first is the case in many languages which have special verb forms for such purposes (often called infinitives). The examples in (1) are from Turkish and Zulu, and the English translations also illustrate this type. (For the sake of clarity, the complement phrase or clause is enclosed in square brackets in all examples of this chapter.)
(1)
a.
Turkish (Wendt 1972: 56)
[Ankara-ya
git-mek]
ist-iyor-um.
Ankara-all
go-inf
want-impf-1sg
'I want to go to Ankara.'
b.
Zulu (Ziervogel et al. 1981: 210)
Ngi-funa
[uku-dlala
na-ye].
1sg-want
inf-play
with-him
'I want to play with him.'
Notionally, there are two predications in these constructions: the predication of wanting, and the predication that expresses the desideratum. The latter is expressed in these examples by a complement phrase or clause that lacks an overt subject. The constructional meaning makes it clear that the notional complement subject is to be understood as being the same as the experiencer argument of 'want' in the main clause.
Alternatively, the complement may look more like a complete clause in that the complement subject is expressed overtly by a pronominal element, generally a simple verbal person-number marker. Verb forms used in such complement clauses are sometimes called subjunctive. The examples in (2a-b) are from Modern Greek and Egyptian Arabic.
(2)
a.
Thél-o
[na
pá-o
s-tin
Athína].
want-1sg
comp
go-1sg
to-art
Athens
'I want to go to Athens.'
b.
Egyptian Arabic (Mitchell 1962: 144)
ʔana
ʕaawiz
[a-kallim
il-mudiir].
I
want
1sg-talk
art-manager
'I want to talk to the manager.'
Literally, the sentence in (2a) can be translated 'I want that I go to Athens'. The suffix -o in the subordinate verb pá-o expresses the subject in much the same way as the pronoun I does in English. When the wanter is a full noun phrase, it is of course not fully repeated in the complement clause, but is represented only by the appropriate pronominal element (e.g. Modern Greek I Rúla thél-i na pá-i [art Roula want-3sg comp go-3sg] 'Roula wants to go', literally 'Roula wants that she go').
Another possibility is that a language has both constructions, one that has an implicit complement subject, and another one that has an overtly expressed subject. An example of this type is Obolo (Delta Cross, Niger-Congo; southern Nigeria):
(3) Obolo (Faraclas 1984: 104, 112)
a.
M´-wèèk
ígêgè
íkpá.
1sg-want
inf.write
letter
'I want to write a letter.'
b.
M´-wèèk
ǹ-gê
íkpá.
1sg-want
1sg-write
letter
'I want to write a letter.'
Finally, quite a few languages do not express the notion of wanting by means of an independent verb (or other lexeme), but by a desiderative marker. This can be a verbal affix, as in (4) from Apurinã (Arawakan; Amazonas, Brazil).
(4) Apurinã (Facundes 2000: 321)
nhi-nhika-ene-ta-ru
1sg-eat-desid-vblz-3m.o
'I wanted to eat it'
Alternatively, the desiderative marker can be an uninflected particle meaning 'want' that cooccurs with a fully inflected verb which expresses the desideratum predication. An example comes from Itzaj (Mayan; Guatemala).
(5) Itzaj (Hofling 2000: 48)
Tak
u-wen-el.
desid
3abs-sleep-incompl
'S/he wants to sleep.'
With desiderative markers, the question of whether the complement subject is implicit or expressed does not arise because there is no complement phrase or clause. The two notional predications of wanting and eating/sleeping are expressed in a single clause in (4-5), and obviously the subject is expressed only once here.
Thus, there are five different language types that are distinguished on the map:
| Go to map | ||
|---|---|---|
| Value | Representation | |
 |
The complement subject is left implicit | 144 |
 |
The complement subject is expressed overtly | 72 |
 |
Both construction types exist | 14 |
 |
'Want' is expressed as a desiderative verbal affix | 45 |
 |
'Want' is expressed as an uninflected desiderative particle | 8 |
| Total: | 283 | |
It should be noted that this chapter only covers constructions involving coreference between the wanter and the subject of the desideratum predication. When these two are not coreferential (as in 'I want Roula to go to Athens'), many languages use a completely different construction.
When the complement subject is left implicit, languages often use a special verb form such as the “infinitives” that were illustrated in (1) and that we also find in many Indo-European languages. Forms called “infinitives” generally lack not only subject-agreement marking, but also tense-marking, and often they have nominal properties such as the possibility of occurring with adpositions. However, some languages do use verb forms that include a tense-marker in implicit-subject complements, for instance Panyjima (Pama-Nyungan; Western Australia).
(6) Panyjima (Dench 1991: 200)
Ngatha
purlpi
[puntha-rta].
I.nom
want
swim-fut
'I want to swim.'
And sometimes the bare stem of the verb is used in such constructions, even in languages with a fair amount of morphology. Example (7) is from Páez (Paezan; Colombia).
[Wakar-ti's-a']
nasa
[wala
uy]
wê-xe-ts-me:-ta'.
white-dat-top
Páez
much
see
want-impf-prog-neg-decl.3pl
'The Páez don't want to see the whites much.'
In languages with little morphology, this is the most common pattern; see example (8) from Maori, which in addition to the verb stem uses a complementizer (this is also the case in English).
(8) Maori (Bauer 1993: 42)
Ka
piirangi
ia
[kia
moohio
ki
te
kootiro
raa].
tns
want
he
comp
know
case
art
girl
that
'He wanted to know that girl.'
In the great majority of cases, an overtly expressed subject in a 'want' complement clause is a person-number affix, like -o in Modern Greek (example 2a) and (ʔ)a- in Egyptian Arabic (example 2b). Languages with obligatory subject pronouns (i.e. non-affixal elements; see Chapter 101) generally leave the subjects of their 'want' complements implicit, but some exceptions to this generalization can be found, for instance in Dagbani (Gur, Niger-Congo; Ghana).
(9) Dagbani (Olawsky 1999: 29)
Doo
maa
daa
bɔri
[ni
o
ŋubi
noo].
man
the
td
want
subord
he
chew
fowl
'The man wanted to eat fowl.'
In quite a few languages, the complement verb is additionally characterized by a special marker on the verb (often called subjunctive or a similar label), e.g. in Abkhaz (Northwest Caucasian; Georgia).
(10) Abkhaz (Hewitt 1979: 33)
[S-abà-co
ħ°a
eylə-y-k'aa-rc]
yə-taxə̀-n.
I-where-go
say
prev-3sg.subj-learn-purp
he-want-fin
'He wanted to learn where I was going.'
Sometimes a language has only very partial expression of the subject in the 'want' complement. Where the partial expression includes person information, it was counted as subject expression. In Tetun (Central Malayo-Polynesian; Timor), verbs have a person-number affix only for the first person singular. This inflection also occurs in 'want' complements, as in (11), so I classified Tetun as a language of the second type, although in main clauses, given the impoverished agreement, independent pronouns are normally required in order to express the subject.
(11) Tetun (van Klinken 1999: 291)
Ha'u
la
k-o'i
[k-ola
ó].
I
neg
1sg-want
1sg-take
you
'I don't want to take (i.e. marry) you.'
However, in Hunzib (Daghestanian; eastern Caucasus), the complement verb only agrees with its intransitive subject in gender and number; there is no person agreement at all, and no agreement with the transitive subject. Hunzib was therefore classified as an implicit-subject language.
In quite a few languages, there is no unique word meaning only 'want', but 'want' is expressed by a verb that also means 'say' and/or 'think'. In some of these cases, one wonders whether the 'want' meaning should not be ascribed to the mood inflection rather than to the main verb, for instance in Maricopa (Yuman; Arizona), where the complement verb takes the optative suffix -ly:
(12) Maricopa (Gordon 1986: 249)
[Kwsede-sh
m-do-ly]
m-aly'ii-m?
doctor-subj
2-be-opt
2-think-q.realis
'Do you want to be a doctor?'
In a few North American languages, the 'want' complement behaves like a direct-speech complement in that the deictic center is the wanter, so that the complement clause has a 1st person subject. This is the case in Slave (Athapaskan; northwestern Canada).
John
[náohzéh]
enįdhę.
John
1sg.opt.hunt
3.want
'John wants to hunt.'
(Literally: 'John wants: 'Let me hunt'.')
Such languages, too, were classified as belonging to the explicit-subject type.
In many cases, the two constructions that exist side by side in languages of this type are quite different in that, for instance, the implicit-subject construction requires a special infinitival form of the verb, as in (3a) from Obolo. But the two may also simply differ by the presence or absence of a subject pronoun. Ju|'hoan (Northern Khoisan; Botswana) comes close to this; in addition, the explicit-subject construction also requires the complementizer ká.
(14) Ju|'hoan (Dickens n.d.: 36)
Mí
kàrè
(ká
mí)
g!à'ámá
shóré.
I
want
that
I
enter
school
'I want to enter the school.'
In most cases, the desiderative verbal affix only means 'want' (as in (4) from Apurinã). However, languages often use more general non-realized tense-mood forms (called "irrealis", "future", etc.) where English would use want, especially when the reference is to a 1st person subject. In general, such general tense-mood forms were not regarded as desideratives, but exceptions were made when the grammar author explicitly says that this mood form is the only way to express 'want', or when such a form is systematically and commonly rendered as 'want', even with 3rd person subjects, as in Tugun (Central Malayo-Polynesian; Maluku, Indonesia).
(15) Tugun (Douglas 1991: 86)
Marr-ala
la
hira.
3pl.irrealis-take
for
them
'They wanted to take them for themselves.'
When a language has several different construction types and one of them involves a desiderative marker, it was classified as belonging to the desiderative type, because this is the most grammaticalized construction type.
Only the explicit-subject type reveals an interesting geographical distribution: It shows a concentration in North America and Mesoamerica, as well as (remarkably, given the little amount of morphology in these languages) in West Africa. Within Europe, the explicit-subject Balkan languages stand out, but they fit well into an eastern Mediterranean/Middle Eastern context.